Cern business continuity overview
This presentation is the property of its rightful owner.
Sponsored Links
1 / 20

CERN Business Continuity Overview PowerPoint PPT Presentation


  • 96 Views
  • Uploaded on
  • Presentation posted in: General

CERN Business Continuity Overview. Wayne Salter HEPiX April 2012. Overview. What is business continuity? What steps can be taken? What is the status at CERN? What are our current plans? Conclusions. What is business continuity?.

Download Presentation

CERN Business Continuity Overview

An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -

Presentation Transcript


Cern business continuity overview

CERN Business Continuity Overview

Wayne Salter

HEPiX April 2012


Overview

Overview

  • What is business continuity?

  • What steps can be taken?

  • What is the status at CERN?

  • What are our current plans?

  • Conclusions

CERN Business Continuity Overview - HEPiX April 2012 - 2


What is business continuity

What is business continuity?

  • Business continuity is the activity performed by an organization to ensure that critical business functions will be available to customers, suppliers, regulators, and other entities that must have access to those functions.

  • Disaster recovery is the process, policies and procedures related to preparing for recovery or continuation of technology infrastructure critical to an organization after a natural or human-induced disaster. Disaster recovery is a subset of business continuity.

CERN Business Continuity Overview - HEPiX April 2012 - 3


What is business continuity1

What is business continuity?

  • Business continuitymeans that services can be provided with an acceptable level of availability (SLA/SLD….).

    • Level of availability will vary between services

  • Disaster recoverymeans that services can be restored after a major incident within an acceptable amount of time.

    • Time will vary between services

CERN Business Continuity Overview - HEPiX April 2012 - 4


Possible types of incident

Possible Types of Incident

  • Deletion or corruption of a file

  • Failure of a disk

  • Failure of a server

  • Failure of network equipment

  • Localised incident over many servers

  • Incident effecting a large part of data centre

  • Malicious action

  • Preventative or corrective maintenance

CERN Business Continuity Overview - HEPiX April 2012 - 5


Other issues to consider

Other Issues to Consider

  • When looking at BC for a particular service are all dependencies properly handled?

    • E.g. we have redundant systems in two centres but with common networking

    • How can this be properly verified?

      • Repeated testing!!

  • What is actually required in terms of BC?

    • Instant failover?

    • Can one accept a human triggered failover?

    • Can one accept a re-instantiation of service?

    • Is load balancing with loss of capacity acceptable?

CERN Business Continuity Overview - HEPiX April 2012 - 6


What are typical steps that can be taken

What are typical steps that can be taken?

  • Disaster recovery

    • Ensure all software, data, etc. require to provide the services is stored in more than one location

    • The hardware infrastructure can be recreated

      • Try to avoid tight coupling between applications and hardware

    • The applications (with associated data) can be re-established

    • Knowledgeable personnel are available and/or good documentation!

CERN Business Continuity Overview - HEPiX April 2012 - 7


What are typical steps that can be taken1

What are typical steps that can be taken?

  • Business continuity

    • Establish adequate redundancy

    • Where applicable use UPS/Diesel systems (also with appropriate redundancy)

    • Service spread over multiple locations in a centre or across centres

      • Servers, racks, PDUs, power feeds, aisles, rooms,

    • Implement redundant services

    • Ensure the capability to recreate services within a ‘short period of time’

      • Infrastructure, H/W, S/W, personnel, …

    • Good cyber security measures

CERN Business Continuity Overview - HEPiX April 2012 - 8


Current measures @ cern

Current Measures @ CERN

  • Disaster recovery

    • Critical data backed up to separate building(s)

    • All tools, scripts and configuration data backed up to a separate building

    • Some services running in local hosting site

    • Able to rebuild services on new hardware

    • But where??

CERN Business Continuity Overview - HEPiX April 2012 - 9


Current measures @ cern1

Current Measures @ CERN

  • Business continuity

    • Redundancy

      • RAID (H/W or S/W)

      • Multiple power feeds to CC

      • Two independent LV power distribution systems in CC; critical (UPS+diesel), physics (UPS only)

      • Many servers have multiple PSUs (all critical servers)

      • In UPS systems themselves

      • HVAC (chillers, AHU and pumps)

      • Multiple rooms (currently 3 at CERN and soon 4)

    • At service level (including core networking equipment)

    • Distributed star points across the site (but all core equipment in CC)

    • Local hosting site

      • Not primarily for BC

      • Also requires networking in CC

CERN Business Continuity Overview - HEPiX April 2012 - 10


Example 1

Example 1

  • For batch:

    • Many batch nodes

    • Spread across the CC

    • On different PDUs

    • On different UPS systems but only on physics (even machines with multiple PSUs)

    • 2 LSF masters with hot standby and these are on critical/physics power

      • Sharing disk-based state via a PES-managed Netappappliance (also on critical/physics power)

CERN Business Continuity Overview - HEPiX April 2012 - 11


Example 2

Example 2

  • For databases the following approach is taken:

    • Data is backed up to TSM (two locations). Plan to move part of the backup to disk.

    • Data is replicated to a different location (local hosting site) using Data Guard (applying the redo log). Allows queries to be made on replica and hence offload production

    • Storage on critical/physics power but only one out of two servers on critical/physics due to lack of sufficient critical power capacity

    • Replication done asynchronously but continuously (most of the time <3 seconds - data streaming)

      • Risk of losing a few seconds of transactions

      • Looking at doing this synchronously without loading the production system

  • Validation of backup, creation of logical copies

    • Reload data from TSM to a clean server and apply all redo logs

    • Make a ‘logical copy’ to a different location and backup to TSM

      • Allows to go back to a particular point in time independently of Oracle version. e.g. financial data stored for 10 years.

      • Allows to verify all backup and recovery scripts plus establish the time to make a recovery

CERN Business Continuity Overview - HEPiX April 2012 - 12


Example 3

Example 3

  • Active Directory

    • Servers at CERN and local hosting site with automatic failover (MS facility)

    • At CERN servers on critical and physics power

    • At local hosting site servers on dual feed

    • All data backed up to TSM (two locations)

  • Mail Servers

    • Some mailboxes at CERN and some at local hosting site with mirror at other site

    • MS log shipping used for synchronization

    • Automated failover (MS Exchange feature)

    • At CERN servers on critical and physics power

    • At local hosting site servers on dual feed

    • All data backed up to TSM (two locations)

CERN Business Continuity Overview - HEPiX April 2012 - 13


Example 4

Example 4

  • DFS File Servers

    • Home directories split between CERN and local hosting site with backup at other site

    • MS DFS Replication used for synchronization

    • Workspace folders only at CERN but with backup synchronized using homemade script

    • At CERN servers physics power only (due to lack of available critical power)

    • At local hosting site servers on dual feed

    • All data backed up to TSM (two locations)

CERN Business Continuity Overview - HEPiX April 2012 - 14


Remaining major risks @ cern

Remaining Major Risks @ CERN

  • Extended computer centre power cut

    • No clean separation between critical and physics equipment and their cooling

  • Major incident in computer centre building, e.g. fire

    • In an electrical or HVAC room

    • In one of the machine rooms

    • In the telecoms room

  • Incident in the fibre room

  • Are all dependencies handled correctly?

    • Can only check with testing on a regular basis

CERN Business Continuity Overview - HEPiX April 2012 - 15


Future plans

Future Plans

  • Securing critical cooling in CC upgrade project

  • Remote hosting

    • Duplicate critical services

  • First classification of services against three BC options:

    • Backup

    • Load balancing

    • Re-installation

  • Second network hub

    • Core networking equipment and connection to distributed star points

    • One of the two links to each of the remote centres

    • Commodity and GEANT Internet connection

    • LHCOPN and LHCONE connections

    • Core networking services, e.g. DNS, DHCP, monitoring, etc.

  • Testing

    • Establish test procedures and perform regular tests

CERN Business Continuity Overview - HEPiX April 2012 - 16


Possible network topology

Possible Network Topology

CERN Business Continuity Overview - HEPiX April 2012 - 17


Second network hub

Second Network Hub

CERN Business Continuity Overview - HEPiX April 2012 - 18

Diagram from Eric


Open issues

Open Issues

  • Full classification of services

    • Backup

    • Load balancing

    • Re-installation

  • Latency

    • Expect 20-30ms latency to remote site

    • A priori no known issues but a number of services will need to be tested

  • QoS

    • Ideas on implementing a first order ‘QoS’ to avoid network problems in case of peak network loads. e.g. splitting TCP and UDP traffic

    • Needs testing

  • Backup strategy at remote site; tape robot, dark disks, cloud storage backup

  • Cost

CERN Business Continuity Overview - HEPiX April 2012 - 19


Conclusions

Conclusions

  • CERN’s approach to business continuity up to now has been largely based on applying a good level of redundancy and backing up data to a separate location

  • However, this still leaves the risk of a major incident in the computer centre

  • With the new remote centre CERN we will look to implement a more complete business continuity strategy

    • Not only by implementing critical systems in both locations

    • But also by creating a second network hub at CERN

CERN Business Continuity Overview - HEPiX April 2012 - 20


  • Login